> how the segment's 'masking image' works?
> I refer to segment function ->Custom -> MaskingImage.
It simply excludes that part of the image from the computation of the
objective function. Therefore, it does not play a role in the
registration, bias field estimation of the parameterisation of the
probability distributions of the various tissues. However, after all
those parameters have been estimated, the estimated parameters are
used to generate some kind of segmentation for the masked out bit.
> Is there another way for improving the segmentation of SPM8?
There are lots of things that are likely to help - but I haven't
implemented them yet.
Best regards,
-John
|